Causal bridging inferences: Validating consistent and inconsistent sequences.

Abstract

Abstract It is proposed that causal bridging inferences must be validated against pertinent knowledge before being accepted by the reader. According to this analysis, understanding each of the consistent sequence, Dorothy poured the on the bonfire, so the bonfire went out, and the inconsistent, Dorothy poured the on the bonfire, but the bonfire grew hotter, invokes the pertinent knowledge, water extinguishes fire. In agreement with this prediction, subjects answered Does extinguish fire? more quickly after reading both consistent and inconsistent sequences than after the control temporal sequence, Dorothy PLACED the BY the bonfire, bonfire grew hotter. However, removing the appropriate conjunction, but, from the inconsistent sequences abolishes its answer time facilitation (Experiment 3). It is proposed that, in the latter case, text ideas feed back to and so qualify pertinent knowledge.Resume Il est propose que les inferences de relations causables doivent etre validees en fonction de connaissances pertinentes avant d'etre acceptees par le lecteur. Selon cette analyse, la comprehension de la sequence logique « Dorothy poured the on the bonfire, so the bonfire went out » et de la sequence illogique « Dorothy poured the on the bonfire, but the bonfire grew hotter » met en oeuvre la connaissance pertinente: « l'eau eteint le feu ». Conformement a cette prediction, les sujets ont repondu a la question « Does extinguish fire? » plus rapidement apres avoir lu les sequences logique et illogique qu'apres avoir lu la sequence de controle temporelle « Dorothy PLACED the BY the bonfire. bonfire grew hotter. » Toutefois la suppression de la conjonction « but », de la sequence illogique, a eu pour effet d'allonger les temps de reponse (experience 3). Dans ce cas - ci, il est donc propose que les idees exprimees dans le texte renvoient a la connaissance pertinente et viennent la modifier.Inference processing has been a central issue in the psychology of language during the past twenty years. One type of inference that has received extensive consideration during this period is the bridging inference (Haviland & Clark, 1974). Bridging inferences serve to establish connections between the current clause and preceding discourse, and accomplish this in one of two closely related ways. First, the bridging inference may simply identify the relationship between two discourse ideas, or propositions. In this regard, the propositional content of Ken stepped on the banana peel, He fell down might be shown as P1 (STEP - ON, TOM, BANANA - PEEL), P2 (FALL - DOWN, TOM) (Kintsch, 1974). However, if the reader detects that the first event caused the second event, then the resulting representation also includes the linking proposition, P3 (CAUSE, P1, P2).Second, a bridging inference may also involve unstated ideas that mediate the explicit ones. That is, to properly understand spy quickly threw his report in the fire, floated up the chimney, one must infer a new idea, such as The report burned to ashes (Singer & Ferreira, 1983).Evidence stemming from the use of both on - line and memory measures has converged on the conclusion that bridging inferences frequently accompany comprehension. On - line measurements have revealed that reading time varies systematically with the semantic distance between a sentence and the antecedent ideas to which it must be bridged (Bloom, Fletcher, van den Broek, Reitz, & Shapiro, 1990; Keenan, Baillet, & Brown, 1984; Myers, Shinjo, & Duffy, 1987; Singer, 1979; see McKoon & Ratcliff, 1980b, for a critique). This result indicates that the reader strives to bridge text ideas during understanding.Memory measures have yielded findings consistent with those stemming from the on - line indices. For example, judgement times for ideas expressing the bridging inferences suggested by a message are similar to the judgement times for explicitly stated ideas. …